CDMA系统中基于SVM的多用户检测算法

Multiuser Detector Based on Support Vector Machine for CDMA System

针对多用户检测的特点将支持向量机(SVM)与多用户检测相结合,构成一种多用户检测算法,应用在异步DS-CDMA系统。该算法能用较少的训练样本获得更佳的分类器,因此它的推广能力较好,且对训练要求的样本数也大大下降。仿真结果表明:该算法的误码性能优于传统匹配滤波检测器以及MMSE多用户检测器。

According to characteristic of multiuser detection, the support vector maehine（SVM） is combined with the multiuser detection, and a multiuser detection algorithm is proposed for asynchronous DS - CDMA system in this paper. The algorithm can obtain the better classifiers by using less training samples, so it leads to more generalization and less training samples than the other learning models. The simulation results show that its performance of error code is better than conventional matched filter and MMSE detectors.